a de nition of industrial organization - nyupages.stern.nyu.edu/~acollard/1 basic io.pdf · i....
TRANSCRIPT
Lecture 1: Introduction, and Review of Basic IOTheory Market Structure Models ∗
The purpose of this lecture is to give a map of theIO theory literature. It will be very fast and non-technical.
What I m assuming is that you have seen bits ofit before. My objective is to give you some senseof what is out there and the issues covered in thebasic IO cannon, so you can go into Tirole or theliterature and find the bits you need when you needthem.
Also, I want to make sure we have some sharedvocabulary.
So this should feel like fast revision.
∗Co-Written with John Asker using Julie Mortimer’sNote
1
I. Introduction
A definition of industrial organization:
“Industrial organization is concerned with the work-ings of markets and industries, in particular the wayfirms compete with each other.”
2
There are two branches of I.O.
• Theories of Markets and Market Structure. Thisbranch treats the firm as a black box and fo-cuses on how firms compete with each other.
• Theories of the Firm. (what the rest of the the-ory lectures are about) This branch investigateswhy some transactions are conducted throughmarkets while others are conducted within afirm. Attempts to look inside the black boxand explain things like firm size, the bound-aries of the firm, and incentive schemes withinthe firm.
3
II. A Brief History of Industrial Organization1. Harvard Tradition (1940 - 1960; Joe Bain)
• Structure-Conduct-Performance
• Structure (i.e., how sellers interact with eachother, buyers, and potential entrants) is a func-tion of number of firms, technology, existingconstraints, products...
• Conduct (i.e., how firms behave in a given mar-ket structure) includes price setting, competi-tion, advertising...
• Performance (i.e., technological efficiency, so-cial efficiency, dynamic efficiency) includes con-sumer surplus, optimal variety, profits, socialwelfare...
• Empirically, use OLS regressions to identify cor-relations (i.e., industry profit = f(concentration))
• Argued that high concentration was bad forconsumers, and paved the way for much anti-trust legislation
• Main weakness: assumption that market struc-ture is exogenous
4
2. Chicago School (1960 - 1980; Robert Bork esp.”The Antitrust Paradox”)
• Firms become big for particular reasons
• Emphasis on price theory (markets work)
• More careful application of econometric tech-niques
• Use different market structures to understanddifferent industry settings or markets
• Monopoly is much more often alleged than con-firmed; entry (or just the threat of entry) isimportant
• When monopoly does exist, it is often transi-tory
5
3. Game Theory (1980 - 1990)
• Emphasis on strategic decision making
• Modeled mathematrically using Nash equilib-rium concept
• Produces a huge proliferation of models whichare often very intuitive theoretically
• However, it is difficult to know which model isthe right one for a real world industry
6
4. New Empirical I.O. (1990 - )
• Combines theory and econometrics in a seriousway
• Sophisticated, computationally intense, com-plex empirical models
• Not all I.O. economists think this way or usethe same methods
• This view of the world is constantly evolving
IMPORTANTLY: Current issue of, say, RAND, willhave papers some of which are NEIO style, someof which are very Game Theoretic and others whichjust try to understand how markets work...
7
Industrial Organisation Journal Rankings
• Top 5 Economics Journals [AER, Economet-rica, JPE, QJE, ReSTud]
• RAND Journal of Economics
• Journal of Industrial Econ., Int.J.I.O., J. Econ.and Mgmt Srategy.
• Review of I.O., J. Law and Econ., J. Reg.Econ, Antitrust Bulletin, Management Science
8
Contemporary Issues in I.O.
• Entry and exit
• Merger analysis
• Product choice: characteristics and location
• Retail Markets
• Price discrimination
• The role of information and monitoring tech-nologies
• Advertising
• Learning-by-doing
• Technological innovation
• R & D spillovers
• Regulation
9
10
III. Perfect Competition Often used as a bench-
mark. As a benchmark, it is very conve-
nient, but one does not often believe the
assumptions of perfect competition in re-
ality.
Definition: An agent is said to be com-
petitive if she assumes or believes that the
market price is given and that her actions
do not influence the market price.
11
P
Q
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZD
�������������������������������������������
S
P ∗
Q∗
(P ∗, Q∗) is market clearing price and quantity.
12
Welfare and Perfect Competition: Under per-fect competition, the first-best outcome is achieved.When you are thinking about evaluating a paper,the question of what is the first-best outcome shouldbe paramount in your minds. Is there Allocative In-efficiency, i.e. do the people with the highest valua-tions get the good? Moreover, is the right quantityproduced in the market?
Examples: Small Business Preference Programs cre-ate both allocative inefficiency and quantity distor-tions. Lack of a Carbon Tax may create quantitydistortion.
13
An Important Feature of Perfect Competition:Under perfect competition, a firm can not affectthe price it faces. Thus, there are no strategic in-teractions between firms. Another way of puttingthis is that each firm’s residual demand curve isflat. This is what we mean when we say a firm is aprice taker, and it implies that the firm’s marginalrevenue equals the price.
MR = P
This is an important characteristic of perfect com-petition. It is not true for any form of imperfectcompetition (monopoly, duopoly or oligopoly).
Importantly, the assumption of perfect competitiondoes not imply anything about large numbers offirms. Some market structures imply that price con-verges to the competitive price as the number offirms gets large. Nevertheless, this is independentof the definition of perfectly competitive behavior.
14
Note: Re Perfect Competition
With locally increasing returns to scale there maybe no market clearing price in perfect competition(depends on demand conditions). Hence ‘NaturalMonoply’
15
Some other micro basics: Price elasticity of demand
The percentage change in demand that results froma one percent change in price:
|εd| =%∆Q
%∆P=
∆Q/Q
∆P/P=
P/Q
∆P/∆Q=dQ
dP·P
Q
εd > 1→ elastic, εd = 1→ unit elasticεd < 1→ inelastic
Of course, this is the market demand, not the resid-ual demand facing an individual firm if the numberof firms (N) is greater than one.
When writing up quantitative results, try to figureout if you can express these as elasticities, ratherthan a number like 23.23.
16
Lecture 2: Monopoly
Definition: A firm is a monopoly if it is the onlysupplier of a product in a market.A monopolist’s demand curve slopes down becausefirm demand equals industry demand.
Four cases:1. Base Case (One price, perishable good, non-IRSCosts).2. Natural Monopoly3. Price Discrimination4. Durable Goods
17
BASE CASE
Monopolist’s Profit Maximization Problem:
maxQ
Π = p(Q)Q− C(Q)
(Choosing P or Q makes no difference because weare selecting a single point on the demand curve.This will not be true when we consider oligopolyproblems.) F.O.C. are:
dΠ
dQ= P (Q) +Q
dP
dQ−dC
dQ= 0
→ P (Q) +QdP
dQ=dC
dQ
→MR = MC
18
P
Q
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZD
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
MR
MC
P ∗
Q∗
(P ∗, Q∗) is profit-maximizing choice.
19
The monopolist chooses output such that the markupequals the inverse of the elasticity of demand:
P (Q)− dC(Q)dQ
P (Q)=−QdP (Q)
dQ
P (Q)
=−QP
dP (Q)
dQ
=1
|εd|> 0
Dead Weight Loss exists because P > MC.
20
2. Natural Monopoly
Definition: Declining average cost over all mean-ingful quantities. The most efficient outcome is fora single firm to produce all output.
Note: IRS is sufficient but not necessary for a nat-ural monopoly.
21
P
Q
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZD
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
MR
MCPd
Qd
AC
P ∗
Q∗
Natural MonopolyDefinition of natural monopoly: declining averagecost over all meaningful quantities. The most ef-ficient outcome is for a single firm to produce alloutput. (For example, public utilities)
22
Monopoly will maximize profits at (P ∗, Q∗). (Pd, Qd)are welfare maximizing, but a firm would make aloss in this case.
Raises the question of the first best regulation for anatural monopoly. Extensive literature on this thatmay touch on in assymetric information / adverseselection section.
23
3. Price Discrimination:
1. We distinguish between three types of price dis-crimination:
• First degree (i.e., perfect price discrimination)
• Second degree (i.e., non-linear pricing such asquantity discounts)
• Third degree (i.e., market segmentation)
2. Price discrimination always increases profits (pro-ducer surplus), but its effects on consumer surplusare ambiguous.
24
P
Q
HHHH
HHHHHH
HHHHHH
HHHHHH
HHHHHH
HHHHHHH
HHHHHH
HHHHHH
HHHHHD1
@@@@@@@@@@@@@@@@@@@@@@@@@@MR1
P ∗1
Q∗1
Price Discrimination by the Monopolist:Market 1
25
P
Q
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
D2
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
MR2
P ∗2
Q∗2
Price Discrimination by the Monopolist:Market 2
26
P
Q
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXD
HHHHHH
HHHHHH
HHHHHH
HHHHHH
HHHHHH
HHHHHH
HHHHHHH
HHH
MR
JJJJJJJJJJJJJ
BBBBBBBBBBBBB
P ∗
Q∗
No Price Discrimination by the Monopolist:A Single Price in Markets 1 and 2
27
Price Discrimination by the Monopolist:Implications for Consumer Surplus in both Markets:
This is an empirical issue essentially. Theory tellsus how to think about the problem, but does notanswer the welfare question (other than to say thatP.C. would do better).
Implications for firm profits? They should go upwith price discrimination. Remember we could al-ways choose optimal p1 and p2, such that p1 = p2 sothe option of price discrimination must do at leastas well.
28
4. Monopolists Selling Durable Goods
Definition of a Durable Good: goods that are boughtonly once in a long time and can be used for a longtime. Examples: cars, houses, land.
(The typical analysis uses perishable or flow goods.)
Since the typical analysis uses perishable goods, wecan use static models to understand pricing deci-sions.
With durable goods, we need to take the futureinto account, and we need dynamic models to un-derstand pricing decisions.
(Note that durability is a relative concept: there aremany different degrees of durability. Definition byU.S. Census: 3 years.)
29
Coase Conjecture.
Consider the example of a monopolist who owns allthe land in the world and wants to sell it at thelargest discounted profit.
In year 1, the monopolist sets a monopoly price andsells half the land. (Think of a linear demand curvewith marginal cost at zero.)
In year 2, the monopolist will want to do the samewith the remaining land, but unless the population isgrowing very quickly, demand for land will be lower.Thus, the monopoly land price in year 2 will belower.
30
Coase conjecture: if consumers do not discounttime too heavily and if consumers expect price tofall in future periods, current demand facing themonopolist will fall, implying that the monopolywill charge a lower price (compared to a perishablegood). (Price is driven to marginal cost “in theblink of an eye.”)
Crucial assumptions:
1. durable good2. demand does not grow quickly over time3. consumers anticipate price cuts
Showing the extent to which this conjecture is truehas lead to a long literature, and much of the gametheoretic literature on bargaining starts here (seethe chapter in Fudenberg and Tirole)
31
Choosing a lower relative level of durability is oneway of solving the problem of consumers’ expecta-tions of future price discounts.
Other ways include:
• Renting (as opposed to selling)
• Planned obsolescence (new car models, newfashions... as long as costs are not too high)
• Capacity constraints (numbered prints)
• Buy-back provisions (not useful if consumerscan damage good, or easily resell)
• Announcements/advertising future prices
32
BASIC OLIGOPOLY MARKET STRUCTURE MOD-ELS
Outline:
• Cournot (Nash-in-quantities)
• Cournot with Sequential Moves
• Bertrand (Nash-in-prices)
• Bertrand with capacity constraints
• Cournot versus Bertrand
• Self-enforcing Collusion
Cournot and Bertrand are sometimes referred toas “conjectural variation” models of firm behavior.However, they reduce to Nash equilibria.
33
1. Cournot (Nash in Quantities)
Cournot wrote in 1838–well before John Nash!
He proposed an oligopoly-analysis method that is(under many conditions) in fact the Nash equilib-rium w.r.t. quantities.
Firms choose. . . production levels.
A. Simple Two-seller game
Cost: TCi(qi) = ciqi
Demand: p(Q) = a− bQ where Q = q1 + q2
34
Define a game:
Players: firms.Action/Strategy set: production levels/quantities.Payoff function: profits, defined:
πi(q1, q2) = p(q1 + q2)qi − TCi(qi)
Now we need an equilibrium concept.
35
{pc, qc1, qc2} is a Cournot equilibrium (“Nash-in-quantities”equilibrium) if:
1. a) given q2 = qc2, qc1 solves maxq1π1(q1, qc2)
b) given q1 = qc1, qc2 solves maxq2π2(qc1, q2)
2. pc = a− b(qc1 + qc2), for pc, qc1, qc2 ≥ 0.
“No firm could increase its profit by changing itsoutput level, given that other firms produced theCournot output levels.”
This is just the Nash Equilibrium concept applied toour game. If you aren’t familiar with this concept,go take a game theory course and then take thiscourse next year.
36
π1(q1, q2) = (a− bq1 − bq2)q1 − c1q1
F.O.C. (for firm 1):
∂π1(q1, q2)
∂q1= a− 2bq1 − bq2 − c1 = 0
Best response is:
q1 = R1(q2) =a− c1
2b−
1
2q2
37
q2
q1
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
R1(q2)
QQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQ
R2(q1)
qc1
qc2
a−c1
2b
a−c2
2b
38
Note both best-response functions are downwardsloping. For each firm: if the rival’s output in-creases, I lower my output level. (i.e., an increasein a rival’s output shifts the residual demand facingmy firm inward, leading me to produce less.)
Now we can compute Cournot equilibrium outputlevels.
qc1 =a− 2c1 + c2
3b
qc2 =a− 2c2 + c1
3b
Equilibrium quantity supplied on the market is
Qc = qc1 + qc2 =2a− c1 − c2
3b
39
We can also find equilibrium price.
pc = a− bQc =a+ c1 + c2
3
What are the pay-off functions in equilibrium?
πci = [(a− bQc)− ci](qci)
= (pc − ci)(qci) = b(qci)2
40
Extending this to N firms.
It’s harder to see the reaction functions, but thestory is exactly the same.
Now each firm maximizes profits according to:
πi(q1, q2, ...qN) = p(Q)qi − TCi(qi)
We would derive the best response function for allN firms. For firm 1,
q1 = R1(q2, ..., qN) =a− c1
2b−
1
2(N∑j=2
qj)
41
We need N of these equations. However, if we as-sume that firms’ costs are the same (TCi(qi) = c∀i),it’s a lot easier. Each firm has the same reactionfunction, which is
qi = Ri(q−i) =a− c
2b−
1
2(N∑j 6=i
qj)
42
Back to the reaction function.
qi = Ri(q−i) =a− c
2b−
1
2(N∑j 6=i
qj)
q =a− c
2b−
1
2(N − 1)q
Thus,
qc =(a− c)
(N + 1)b
Qc =N(a− c)(N + 1)b
Now it is straightforward to solve for pc (the marketprice) and πci (profits for each firm).
43
Sanity checks. . .
Do we get the monopoly result for N = 1?Do we get the duopoly result for N = 2?What is the Cournot solution for N =∞?
Take the limit as N →∞ for qc and Qc and pc. Theyare:
limN→∞
qc = 0
limN→∞
Qc =a− cb
limN→∞
pc = c
44
Under the assumption of Cournot competition, mar-ket supply approaches the competitive supply asN →∞.
Note that market supply depends on the slope andintercept of demand, and the (common) marginalcost. Individual firms’ output levels approach zeroas N →∞.
45
Take a look at the results when firms have differentlevels of marginal cost.
Let marginal cost of firm i be ci. The F.O.C. isonly a slight modification from before.
πi = [(a− bqi − bq−i)− ci](qi)
∂πi
∂qi= a− 2bq∗i − b
∑i 6=j
q∗j − ci = 0
46
Rewrite the F.O.C. as
a− bQ∗ − bq∗i = ci
Substitute in for price:
p− ci = bq∗iQ∗
Q∗
Rearrange...p− cip
=bq∗iQ∗
Q∗
p
p− cip
=−∂p∂Q
q∗iQ∗
Q∗
p
The term q∗iQ∗
is the market share of firm i. Denote
this simply as si.
47
We now have the inverse elasticity rule of Cournotoligopoly:
p− cip
=si
−εd
And note that for perfectly competitive markets:
p− cip
=0
−εd
And for monpolistic markets:
p− cip
=1
−εd
48
Cournot with Sequential Moves:
We could also think about this in a game where firm1 moves first, firm 2 moves second, etc. We callthis a leader-follower market structure, or a Stack-elberg game. The sequential moves game is a veryreduced-form way to think about situations with in-cumbent firms and potential entrants.
• Work backwards: Suppose firm 1 sets outputlevel to q1. What would firm 2 do?
• R2(q1) = a−c2b− 1
2q1
• Firm 1 can figure this out. What will Firm 1do in response?
• maxq1 p(q1 +R2(q1))q1 − cq1
49
What’s different between this profit function and aCournot profit function?
• Turns out leader output level is: qS1 = a−c2b
=32qC.
• Follower output level is: qS2 = a−c4b
= 34qC
• Equilibrium price is lower than Cournot, outputis larger than Cournot – hence more consumersurplus.
• First-mover advantage: “leader” makes moreprofit than “follower”. “Leader” is better offthan in Cournot.
50
Bertrand (“Nash in Prices”)
When do you think price setting makes more sensethan setting quantity?In general, economists may believe that different as-sumptions hold for different settings. Then we haveto argue about which one is more consistent withthe data.
Bertrand reviewed Cournot’s work 45 years later.
Go through a two-firm example again. Now firmsset prices. We need two assumptions.
1. Consumers always purchase from the cheapestseller (recall defn of homogeneous goods).2. If two sellers charge the same price, consumersare split 50/50.
51
The second assumption is that qi takes the values:0 if pi > a0 if pi > pja−p2b
if pi = pj = p < aa−pib
if pi < min(a, pj)
Do you see a difference between Cournot and Bertrand?
Bertrand has an important discontinuity in the game(more specifically, there is a discontinuity in the pay-off functions.)
Solving for the equilibrium required us to say some-thing about costs.
52
1. If costs are the same, a Bertrand equilibrium isprice = marginal cost, with quantity supplied on themarket equal to the perfectly competitive outcome(equally split between the two firms).
2. If costs differ, (say firm 1 has cost = c1 wherec1 < c2), then the firm with the lower cost chargesp1 = c2 − ε, firm 2 sells zero quantity, and firm 1sells quantity given by qb1 = (a−c2+ε)
b.
Intuition:
If costs are the same, undercutting reduces price tomarginal cost. If costs differ, undercutting reducesprice to “just below” the cost of the high-cost firm.
53
An important aspect of bertrand is that equilibriummay not exist if marginal costs are not constant.This has lead to various models, the most sucess-ful of which is the ‘Supply function equilibrium’ ofKlemperer and Meyer (1989) Econometrica.
54
Bertrand under capacity constraints (keep the as-sumption that costs are the same):
Note that if firms choose capacity then prices, youcan get outcomes more like Cournot. This dependscrucially on the rationing protocol via which con-sumer match to transactions. Tirole is quite goodon this. This model is called Kreps-Scheinkman(1983).
55
Comparison of Equilibrium in Models Covered SoFar:Monopoly (M), Cournot (C), Stackelberg (S), Per-fect competition and Bertrand (PC)
CSM < CSC < CSS < CSPC
PSM > PSC > PSS > PSPC = 0
DWLM > DWLC > DWLS > DWLPC = 0
56
Differentiated Products
We now finally drop the assumption that firms offerhomogeneous products.
Differentiated product models are among the mostrealistic and useful of all models in IO.
If you understand the basic elements of productdifferentiation theories, then you should have anawareness of the economics underlying:
1. product placement2. niche markets3. product design to target certain types of con-sumers4. brand proliferation, etc.
57
There are two aspects of product differentiation:1. Horizontal differentiation: if all products werethe same price, consumers disagree on which prod-uct is most preferredEg. films, cars, clothes, books, cereals, ice-creamflavors, Starbucks (by geographic location), ...
58
2. Vertical differentiation: if all products were thesame price, all consumers agree on the preferenceranking of products, but differ in their willingness topay for the top ranked versus lesser ranked productsEg. computers, airline tickets, different quantities,car packages, ...
59
Differentiated Products Betrand Equilibrium
Now we examine a pricing equilibrium with differ-entiated products. It should be no surprise that wesolve for a Nash-in-prices equilibrium.
There a several cute models of pricing in differen-tiated products Bertrand where people set up thecross-elasticities so that there is enough structureand symmetry to give analytical results. Basically,these models are not that useful for applied workhowever. Which is a shame because differentiatedBertand is hyper-helpful way to conceptualize manyindustries.
Key things to realize: The discontinuity in homoge-nous products Bertrand goes away.
Things look more like Cournot (in the sense of pos-itive margins etc)
(pi − c)di(pi, pj)
First order condition as expected. The extent towhich best response slopes away from 45 degreeline depends on cross-easticities (see next slide fordiagram of what I mean). This in turn affects theextent to which things start to look somewhat sim-ilar to cournot.
60
Best Response Function, 2 firms, Bertrand
p1
p2����������������������������
pb2
R2(p1)
����
���
���
���
���
����
���
����
R1(p2)
a2b
61
“Industry structure” refers to the number of firms inan industry and the size distribution of those firms,among other things.
Many policy issues revolve around the concentrationand number of firms in a particular industry and howcompetitive we think the industry is.
i.e., Why do some markets have a small number offirms while other markets have a large number offirms?
We’ve already seen that the number of firms in anindustry can have a big impact on the supply equi-librium in the market, but we have not discussedhow the number of firms might arise endogenouslyin the industry.
62
Notation:N is number of firmsQ is total industry outputqi is output of firm i, so Q =
∑Ni=1 qi
si is % share of firm i (ie., si = 100 ∗ qiQ
)
Two Measures of Concentration:
1) Four-firm Concentration Ratio:I4 =
∑4i=1 si
Examples (1992, 2-digit SIC classifications):Tobacco: 91.8Furniture: 29.3Department stores: 53.8Legal services: 1.4
Problems with this or any other “linear” measure:No difference between (80,2,2,2) and (20,20,20,26)(ie, can’t distinguish concentration between the top4.)
63
2) Herfindahl-Hirshman Index (HHI)
IHH =∑N
i=1(si)2
Solves the linearity problem. Notice:
Monopoly: IHH = 10,00010 identical firms: IHH = 1,000
DOJ Merger Guidelines use IHH. If IHH > 1,800,then mergers come under scrutiny.
64
Problems with Using Concentration Measures Gen-erally:
Market Definition: How should we define marketsand industries when products are differentiated?
Does not get to central issue about how firms be-have. For example, consider the case with 2 equal-sized firms, but in one industry, they compete Cournot,and in another, they compete Bertrand. The IHH =2,500, but we only worry about the Cournot indus-try.
This is why sticking HHIs into a cross-industry re-gression to take care of ‘competition’ is not goingto get a warm welcome from an IO guy.
65
Entry
We will focus on entry in two different contexts.
1. Non-strategic Effects on Entry (Entry Barriers)
These are features of firms’ costs or productiontechnologies that affect how many firms can effi-ciently serve a market. For example, natural mo-nopolies, generally the m.e.s. compared to the mar-ket size. Other features could include absolute costadvantages, regulatory restrictions (licensing, etc.),capital requirements...
2. Strategic Effects on Entry (Entry Deterrence)
These are costs of entry borne by entrants (or po-tential entrants) that are a result of strategic behav-ior by incumbent firms. Some examples are capac-ity commitment, spatial preemption, limit pricing,long-term contracts, and other actions that an in-cumbent firm might take in the presence of an entrythreat that he would not take otherwise. (Possiblyalso tying or other arrangements that have beendiscussed in the Microsoft case.)
66
Usually barriers to entry can be expressed in termsof sunk costs.
1. Exogenous Sunk Costs
Costs that are embedded in the underlying condi-tions of the market and not determined by firms’actions.
2. Endogenous Sunk Costs
Conditions and strategic actions that firms withinthe industry can change. These are entry costs forwhich the firm has choice over how large they willbe.
67
Examples:
1. Exogenous Sunk Costs or Barriers to Entry:Capital requirementsScale economiesAbsolute cost advantagesAsset specificityRegulatory restrictions (licensing)
2. Endogenous Sunk Costs or Barriers to Entry:R&DPatentsExcess capacityControl over strategic resourcesContractsAdvertising?
68
Why advertising?
There is debate about this. The argument for think-ing of advertising as a barrier to entry is:
It does not depend on the level of outputThe effect of advertising is to increases consumers’willingness to pay for that productThere are no spillovers that benefit other firms (usu-ally)All consumers in the market are affected
The opposing side of the debate says:
If capital markets are efficient, a new entrant willjust borrow the money necessary to do his own (pos-sibly higher) level of advertising.
Both sides have a point: perhaps it depends on themarket. (i.e., pepsi and coke vs. kitchen appliances)
69
Usual form of these types of models
Timing of events:
1) Each firm decides whether to enter an indus-try/market2) There are some exogenous sunk costs (ie., theacquisition of a plant of minimum efficient size)3) Each firm then chooses some endogenous sunkcosts (ie., advertising or R&D)4) Finally, firms in the market engage in some formof competition
Stages 3 and 4 can be pretty complicated, depend-ing on the model. Eg Reputation models (gang of4 etc), some of Whinston’s exclusion papers etc
70
Lecture 11: Vertical Control (Part 1)
Manufacturers rarely supply final consumers directly.Instead, most industries are vertically separated. Weoften refer to firms “vertically-separated markets”as upstream and downstream firms.
In these settings, downstream firms are the cus-tomers of the upstream firms, and many of the is-sues that we have reviewed already still apply. Forexample, the upstream firm may want to price dis-criminate across the downstream firms.
71
However, things can also get more complicated invertical relationships between firms. In particular,downstream firms often do not simply consume thegood, but typically make further decisions regardingthe product.
Examples of activites of downstream firms:
1) determination of final price2) promotional effort3) placement of product on store shelves4) promotion and placement of competing products5) technological inputs
72
Unlike the consumption activities of final consumers,the activities of the downstream firms may affectthe profits of the upstream firm. This is why up-stream firms care about the activities of the down-stream firms, and why we study vertical control/restraintsbetween firms in these settings. IO research hastended to focus on the incentives for vertical controlwhen the market for the intermediate good is im-perfectly competitive because this is where antitrustconcerns are most apparent. This leaves fairly openthe question of why they do it in other contexts(although read Williamson, ‘The Economic Institu-tions of Capitalism’)
A common benchmark for what firms can achievethrough vertical control is the “vertically integratedprofit.” This is the maximum industry or aggregate(manufacturer plus retailer) profit. If firms use ver-tical restraints efficiently, they should achieve thevertically integrated profit.
73
Often vertical restraints used by firms in vertically-separated markets are grouped into 5 classes:
• Exclusive Territories: a dealer/ distributor/ retailer isassigned a (usually geographic) territory by the manu-facturer/ upstream firm and given monopoly rights tosell in that area.
• Exclusive Dealing: a dealer/ distributor/ retailer is notallowed to carry the brands of a competing upstreamfirm.
• “Full-line forcing”: a dealer is committed to sell all vari-eties of a manufacturer’s products rather than a limitedselection. (the upstream firm ties all products when sell-ing to the downstream firm).
• Resale Price Maintenance: a dealer commits to a retailprice or a range of retail prices for the product. Thiscan take the form of either minimum resale price main-tenance or maximum resale price maintenance.
• Contractual arrangements: upstream and downstreamfirms write contracts to provide greater flexibility in thetransfer of the product. Profit sharing and revenue shar-ing are the most common, which we’ll see soon. Also,quantity forcing and quantity rationing and franchise fees.
74
The typical outline of vertical control is as follows:
1) Basic Framework
2) The need for control because of externalities be-tween downstream and upstream firms, or amongdownstream firms themselves.
3) Intrabrand competition
4) Interbrand competition
Think of exclusive territories as a form of verticalcontrol to restrain intrabrand competition, and ex-clusive dealing as a way of restraining interbrandcompetition.
75
Basic Framework: Double Marginalization
Simple model: homogeneous good with (inverse)linear demand given by
p = a−Q
(ie. I’m keeping things super simple to show youflavor fast)
Suppose we have a monopolistic manufacturer andwe have given exclusive rights to a dealer to sell theproduct of the manufacturer, so both the upstreamand downstream firms are monopolistic. The down-stream firm has marginal cost of selling the productof d which is equal to the wholesale cost of pur-chasing the product from the manufacturer, andthe manufacturer has marginal cost of producingthe good equal to c.
76
Dealer maximizes his profit given by
πd = p(Q)Q− dQ = (a−Q)Q− dQ
F.O.C.:
∂πd
∂Q= 0 = a− 2Q− d
Q∗ =a− d
2p∗ =
a+ d
2πd =
(a− d)2
4
Now, how should the upstream firm set d?
Check: what are the strategies of the two playersin this game? What does each firm choose?
77
Manufacturer maximizes profit given by
πm = (d− c)Q = (d− c)a− d
2
F.O.C.:
∂πm
∂d= 0 = a− 2d+ c
d∗ =a+ c
2πm =
(a− c)2
8
Note that we can now substitute into the dealer’ssolutions (for d) and get:
Q∗ =a− c
4p∗ =
3a+ c
4πd =
(a− c)2
16
78
Results:
1. The manufacturer earns a higher profit than thedealer
2. The manufacturer could earn a higher profit ifhe does the selling himself. Total industry profitin this case is lower than the vertically integratedprofit. Shown here:
πV I =(a− c)2
4> (πd + πm) =
3(a− c)2
16
The presence of two markups screws things up forthe firms. This basic fact is called:double-monopoly markup problem, successive mo-nopolies problem, or double marginalization.
79
P
Q
ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZDr
JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
MRr = Dw
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBMRw
MC
Pw
Pr
80
As mentioned earlier, there are many ways aroundthese problems, including RPM, contracts, etc. Thereare also other problems that arise, and sometimeswe might even create a successive monopoly prob-lem in order to solve other incentive problems in thevertical channel.
81
Resale Price Maintenance:
Requires retailers to maintain a minimum price, amaximum price, or a fixed price. Examples: Win-dows 98, Windows XP, books, many many retailproducts.
Two goals:
1) Partially solve the double marginalization prob-lem
2) Can induce dealers or retailers to allocate re-sources for promoting the product, or exerting otherforms of effort in distributing the product. (Exam-ples: perfume, Coors beer)
82
Consider the example of promotions or advertising.Assume (inverse) demand is given by
p =√A−Q
The manufacturer sells to two dealers who competein price. Denote the wholesale price as d and adver-tising expenditures as A1 and A2, where A = A1+A2.
First result:
For any given d, no dealer will engage in adverits-ing and demand would shrink to zero, with no sales.
83
Why?
Firms compete in price, and they sell a homoge-neous product. What does p equal in this case??
84
What can Resale Price Maintenance do?
Minimum Resale Price Maintenance: p = pf ≥ d
Now demand is
Q =√
(A1 +A2)− pf
Assume that quantity demanded is split evenly be-tween the two retailers. The only strategic variablefor the retailers is A. Thus, writing profits as afunction of A and finding the F.O.C. yields:
πi =
√(Ai +Aj)− pf
2(pf − d)−Ai
85
F.O.C.:
0 =∂πi
∂Ai=
pf − d4√
(Ai +Aj)− 1
Note that we can only identify the sum of A1 + A2
and not A1 and A2 individually. But the idea isthat retailers will compete on promotion now. Aslong as pf > d then at least one retailer has anincentive to advertise, and the total dollars spenton ads increases with the markup.
86
Note that one problem in the last example was thatcompetition between the retailers initially resultedin too much competition downstream, so that firmscould not afford to advertise as a vertically-integratedfirm would choose to do.
One way around that: Exclusive Territories or “Ter-ritorial Dealerships”
87
Legal Issues
There are a lot of ambiguities in legal treatmentof vertical contracts.
–Until 1970s,–RPM and E. Territories were per se ille-
gal under Sherman Act.
–But many states passed fair trade laws that wereinterpreted to cover some of these cases.
Thus, although price fixing remains per se illegal,it’s not always applied in vertical settings b/c itconflicts with free–trade notions between mfgs andtheir distributors.
–Non-price issues have been generally accepted tobe ok by the courts
• Exclusive territories
• Refusal to deal
• . . . Foreclosure?
88